Long-Run Growth Forecasting (eBook)

The author works for a commercial bank and has been the lead researcher in the bank's project called "Global Growth Centres 2020".

Acknowledgement 7
Contents 9
List of variables and coe cients 13
List of abbreviations 15
1 The importance of long-run growth analysis 16
1.1 Frequent forecast failures 16
1.2 Strong demand - but little supply 18
1.3 Plan of work 21
1.3.1 Choosing a sensible theoretical model 21
1.3.2 Choosing the best econometric technique 23
2 Assessment of growth theories 24
2.1 The search for a dynamic model 24
2.2 The basic neoclassical model 25
2.2.1 Application in cross-country analysis 27
2.3 Focus on convergence 28
2.3.1 Tests for conditional convergence 30
2.4 Models with deeper insights 31
2.4.1 Including human capital (Lucas) 32
2.4.2 Modeling barriers to riches (Parente & Prescott)
2.5 Opening the theories further 34
2.5.1 Models with scale effects 35
2.5.2 Evolutionary models of growth 36
2.5.3 Open-system models 39
2.6 General critique of the standard approach 40
2.6.1 Production function cannot be estimated 40
2.6.2 Aggregate production function does not exist 43
2.6.3 The concept of TFP is not helpful 44
2.6.4 Beyond neoclassical economics 44
2.7 The augmented Kaldor model 45
3 The dependent variable: GDP growth 50
3.1 Choosing the appropriate data source 51
4 Labor input 58
4.1 Population growth is endogenous 58
4.2 Hours worked per capita are important 61
5 Physical capital 66
5.1 Measuring capital accumulation 67
5.1.1 Investment and changes in capital stocks 67
5.1.2 Di erent databases - different investment ratios 68
5.1.3 Capital stocks from perpetual inventory 69
5.2 Main insights on capital accumulation 71
5.2.1 Investment ratios are not constant 71
5.2.2 Investment ratios do not differ much across countries 71
5.2.3 Investment ratios are not proportional to changes in the capital stock 75
5.2.4 Investment ratios are not proportional to levels of the capital stock 76
5.2.6 Capital accumulation is not exogenous 78
5.3 Proper modeling of capital accumulation 78
6 Human capital 82
6.1 Micro- and macroeconomic theory 84
6.1.1 Microeconomic analysis: labor economics 85
6.1.2 Macroeconomic models with different conclusions 86
6.2 Measures and empirical analysis 89
6.2.1 Best measure: years of education 90
7 Openness 96
7.1 Theory: higher efficiency 98
7.1.1 Extent of the market and specialization 99
7.1.2 Good macro polices and more competition 99
7.1.3 Additional influences of trade on income 101
7.2 Measuring openness 101
7.2.1 Black market premium and tariffs 102
7.2.2 The openness dummy 102
7.2.3 Best measure: adjusted trade share 103
8 Spatial linkages 110
8.1 Spatial economics - location matters 112
8.1.1 Absolute location: latitude and climate 112
8.1.2 Relative location: rich neighbors 112
8.2 Constructing spatial GDP 113
8.3 Sum: Spatial linkages not much help 116
9 Other determinants of GDP 118
10 The theory of forecasting 120
10.1 The benefits of forecast experiments 121
10.2 The characteristics of good forecasts 121
10.3 Intercept correction and forecast combination 124
11 The evolution of growth empirics 128
11.1 Still widely used: cross-section 129
11.2 Weaknesses of cross-section regressions 131
11.2.1 Same production function assumed 132
11.2.2 Long-run growth path assumed to be constant and the same across countries 132
11.2.3 Same pace of conditional convergence assumed 133
11.2.4 Errors are assumed uncorrelated with the explanatory variables 133
11.2.5 Right-hand side variables assumed exogenous 133
11.2.6 In sum: many assumptions are violated 134
11.3 The climax of cross-section 134
11.4 Advantages of panel techniques 136
11.4.1 Initial technology can differ across countries 138
11.4.2 Dealing with endogeneity bias 138
11.4.3 Addressing lagged dependent bias 139
11.4.4 Modeling heterogeneous technological progress 140
11.4.5 Summary of results from panel regressions 140
11.5 Non-stationary panel techniques 141
11.5.1 Pooled mean group technique 141
11.5.2 Testing unit roots and cointegration in panels 143
11.5.3 Panel unit root tests 144
11.5.4 Panel cointegration tests 146
11.6 A two-stage estimation method 149
Stage 1: Cointegrating relationships. 150
Stage 2: Stationary models. 150
12 Estimation results 152
12.1 Correlation analysis 152
12.2 Panel unit root tests 154
12.3 Panel cointegration test 157
12.4 The short-run forecasting models 161
13 Forecast competitions and 2006-2020 forecasts 166
13.1 Forecast competition 2001-2005 166
13.2 Forecast combination 169
13.3 Forecast competition 1996-2005 170
13.4 Forecasts for 2006-2020 170
13.5 Other long-run forecasting models 175
14 Conclusion and outlook 178
List of figures 180
List of tables 182
References 184